Lung cancer is a major public health issue and the leading cause of cancer-related deaths worldwide. Being that the clinical outcome of lung cancer patients is still poor (despite advancements in treatments, such as surgery, chemotherapy and radiotherapy), the need to detect this disorder at an early stage is certainly great. There is no doubt that improvements of the current diagnostic methods (i.e., computed tomography and chest X-ray) and efforts aimed at the identification of effective biomarkers with a systemic relevance to this disease would be expected to provide a substantial contribution toward increasing the survival rate of lung cancer patients. In this respect, clinical proteomics has emerged as an approach with an eye toward looking further ahead into disease mechanisms in which proteins play a major role.
3.1. Cell Cultures
If in vitro
culture-based cells are the most widely used model system for the identification of potential biomarkers for the early detection and prognosis of lung cancer, iTRAQ (isobaric tags for relative and absolute quantitation) is one of the most interesting strategies for the measurement of protein expression level [25
With this method, the tryptic peptides obtained from the sample under investigation are labeled with a tag that is designed to fragment in MS/MS in such a way that mass information from one of the molecule regions (the “reporter” section) can provide quantitative information concerning the relative amount of the peptide in the sample [30
This versatile approach was applied (i) by Keshamouni et al.
] to human lung adenocarcinoma cell line A549; (ii) by Eriksson et al.
] to a small lung cancer cell line, sensitive to the cytotoxic drug, doxorubicin (H69), and to its isogenic doxorubicin-resistant parental cell line (H69AR); (iii) by Chenau et al.
] to a human non-small lung adenocarcinoma cell line of bronchoalveolar origin (H358) with a homozygous deletion of p53; (iv) by Chiu et al.
] to investigate lung cancer cell lines with different invasive and metastatic capabilities (CL1-0 and CL1-5 cells); and (v) by Haura et al.
] for the identification of a protein in a system in which it would not be observed, because of its low-abundance, due to sequestration by an endogenous bait protein in the sample and/or suppression by the high quantity of the bait.
Works on tumor progression and metastasis, aimed at investigating the role of transforming growth factor-β (TGF-β) in inducing the epithelial to mesenchymal transition (EMT) of epithelial cells, demonstrated, for the first time, that this process in A549 cells results in a migratory and invasive phenotype [25
In fact, a quantitative differential analysis on TGF-β treated and untreated cells revealed the overexpression, in the former group, of transglutaminase 2 and β1 integrin. This finding was of particular interest, since transglutaminase 2 is characterized by a critical regulatory activity, being implicated in the transformation of latent to active TGF-β [32
]. In its turn, the upregulation of β1 integrin was previously found to be associated with increased invasion of human lung, breast and ovarian cancer cells [34
] and to be also required for TGF-β-induced EMT in mammary epithelial cells [36
The study aimed at demonstrating the primary resistant mechanism to anthracyclines in tumor cells, evidencing the role of a few key-proteins [26
]. In particular, Serca 2, a Ca2+
pump with ten transmembrane segments located in the endoplasmic reticulum (ER) membrane and responsible for pumping Ca2+
into the ER lumen, was under-expressed in the cell line resistant to doxorubicin. The Ca2+
levels in the ER were previously shown to play an important role in the cell ability to undergo apoptosis [37
]. On the basis of these findings, it was hypothesized that the downregulation of the enzyme responsible for the influx of Ca2+
into the ER contributes to the resistance phenotype of H69AR cells. Further, the finding of plectin, vimentin and moesin among the over-expressed proteins in the resistant cell line suggests a more rigid structure of these cells, making them less sensitive to environmental stress.
Particularly intriguing were the results obtained from the quantitative secretome analysis performed on H358 cells. This is a human non-small lung adenocarcinoma cell line of bronchoalveolar origin with a homozygous deletion of p53, a major tumor suppressor protein very frequently mutated and/or inactivated in cancer cells [38
]. This proteomic analysis evidenced that the p53 deficient cells overexpressed a protein belonging to the fibroblast growth factor family (FGF-19) that binds the FGF-4 receptor only. This finding was very interesting, since this protein was previously shown to be overexpressed in many tumors and to be implicated in tumor progression [39
Upon investigating lung cancer cell lines with different invasive and metastatic capabilities (CL1-0 and CL1-5 cells), retinal dehydrogenase I (ALIA1), peroxiredoxin-I (PRDX1) and nidogen-1 (NID-1) were found to be higher in CL1-0 and collagen alpha-1 (VI) chain (COL6A1), metalloprotease inhibitor 1 (TIMP1), urokinase-type plasminogen activator (uPA) and alpha-1-antitrypsin (AAT) were higher in CL1-5 [28
]. The elevation of the TIMP-1 concentration was previously found to promote cancer progression, enhancing the proliferation of endothelium and angiogenesis [40
], and the µPA system of plasminogen activation was correlated with invasion and metastasis [41
]. On the other hand, elevated serum levels of AAT have also been observed in numerous cases of malignant diseases and different cancers [42
]. In its turn, COL6A1, a major component of extracellular matrix (ECM) involved in the organization of fibronectin and other types of collagen (including types I and IV), was shown to be implicated in cell migration and differentiation [44
]. On the basis of these data, a pathway analysis that showed how COL6A1 is directly regulated by the proteolysis of metalloproteases (MMP-1 and MMP-7) in ECM could be performed. These results also proved the metastatic abilities of COL6A1, a finding that had never been previously reported.
The versatility of iTRAQ reagents was successfully demonstrated by applying it to streptavidin-hemagglutinin (SH)-tagged versions of epidermal growth factor receptor (EGFR), expressed in two cancer cell lines (HCC827 and PC9 cells) via retroviral transduction, with the aim of identifying the EGFR core protein complexes [29
]. It demonstrated that ubiquitin-associated and SH3 domain-containing protein B (UBS3B), a member of the EGFR core complex, was indeed present in both cell lines, although, due to its markedly reduced quantities, it could not be detected by standard procedures.
To identify the phosphorylation sites in human non-small lung cancer (NSCLC) cells, Zhang et al.
] focused primarily on the development of a new method in this area. Their strategy, consisting in a combination of protein immunoprecipitation and nano-LC-MS/MS, allowed for the identification of 30 phosphorylation sites, including 12 tyrosines (pY), 12 serines (pS) and six threonines (pT). In addition, upon scanning phosphorylation of EGFR across 31 lung cancer cell lines, they observed that higher stoichiometry of three phosphorylation sites was statistically correlated with EGFR sensitizing mutations. Similarly, higher amounts of phosphorylation on three sites correlated with the sensitivity of erlotinib, an inhibitor of the tyrosine kinase receptor. Five phosphorylation sites on EGFR from the HCC827 cell line, which were inhibited by erlotinib in a concentration-dependent manner, were also identified.
The characterization of phosphorylated proteins is indeed essential for the understanding of lung cancer development, since phosphorylation is a key process in tumor progression for many diseases. In this context, a significant implementation of a human lung cancer phosphoproteome profile was provided by a number of recent articles [46
]. The first large-scale survey of tyrosine kinase activity in lung cancer was that performed by Rikova et al.
], who characterized tyrosine kinase signaling across 41 NSCLC cell lines and over 150 NSCLC tumors. Their approach was particularly interesting, because it provided insight into cancer biology, not available at the chromosomal and transcriptional levels. Known oncogenic kinases (such as EGFR and c-Met), as well as novel anaplastic lymphoma receptor tyrosine kinase (ALK) and proto-oncogene tyrosine-protein kinase ROS (reactive oxygen species) fusion proteins were identified. In addition, other activated tyrosine kinases, such as platelet-derived growth factor receptor α
) and discoidin domain receptor tyrosine kinase 1 (DDR1), not previously shown in the genesis of NSCLC, were also identified.
The two cell lines with different metastatic abilities (CL1-0 and CL1-5) have been investigated also by Wu et al.
] with the aim of identifying new tyrosine phosphorylation sites involved in the metastatic process. Based on their label-free quantitative approach, a total of 36 P-Tyr (30 P-Tyr with higher levels in CL1-5 cells and six P-Tyr with higher levels in CL1-0 cells) were considered to be associated with metastasis. Seven of these proteins appeared of great interest, because they had never been reported before to be associated with lung cancer metastasis. A protein-protein interaction network analysis of these altered proteins revealed that 11 proteins were linked to a network containing EGFR, tyrosine-protein kinase (c-Src), transcription factor Myc (c-Myc) and signal transducer and activator of transcription protein (STAT), which are known to be related to lung cancer metastasis. Yu et al.
] focused on the large-scale study of human lung cancer A549 cell phosphoproteome that had never been analyzed before. Interestingly, most (68%) of the 337 phosphorylation sites (on 181 phosphoproteins) identified in this work appeared to be novel, no matches being found with the known sites contained in the lung cancer cell database. From among these proteins, Yes-associated protein 1 (YAP1), phosphorylated at Ser-127, was observed in three different cell lines. The fact that its expression resulted in clearly decreased in cancer cells and tissues as compared to controls suggested YAP1 to be a promising lung adenocarcinoma biomarker.
Two additional recent articles [49
] focus on the proteomics of A549 cells. Luo et al.
] analyzed the secretome of this cell line using an approach in which intracellular contaminations were minimized, as assessed by the fact that more than 85% of proteins were accepted as being secretory and around 77% as being extracellular or membrane-bound. The list of qualified secretome included cathepsin D, a candidate that had already emerged as a potential lung cancer biomarker from the analysis of M-BE (a SV40T-transformed human bronchial epithelial cell line) cell secretome [51
]. Moreover, findings that had never been reported before concerned C4b-binding protein (C4BP), a protein that controls the classical pathway of complement activation. It was found to exhibit a strong power of discrimination between normal and lung cancer sera, and C4BP serum levels were associated with tumor staging.
Sitek et al.
] described a label-free differential proteomic analysis performed on the A549 cell line, treated or not with TGF-β. This approach allowed for the detection of 202 differentially expressed proteins.
A large-scale comparative molecular mapping of microvascular endothelium in human NSCLC tissues and adjacent-normal lung tissues was performed by Park et al.
] to identify lung cancer-related endothelial cell (EC)-selective proteins. This material was isolated from five patients with lung cancer by using CD31 antibody, an endothelial marker widely used to enrich ECs in various tissues [53
]. The potential EC-enriched candidate biomarkers included nicotinamide adenine dinucleotide NADH dehydrogenase (ubiquinone), 1 alpha subcomplex 5 (NDUFA5), peroxiredoxin 4 (PRDX4) and thymopoietin (TMPO), which were highly expressed in cancer cells compared to normal ones. Furthermore, COPG, a subunit of the heptameric protein complex coatomer protein I (COP), was found to be highly and specifically expressed in cancer-derived ECs.
Phosphoproteome was also investigated by Rodriguez-Ulloa et al.
] under a different point of view. They performed, for the first time, a proteomic analysis on NSCLC treated with proapoptotic peptide-based drug (CIGB-300). This is a pro-apoptotic peptide-based drug that abrogates the phosphorylation mediated by casein kinase 2 (CK2), a protein overexpressed in a variety of solid and hematopoietic tumors [55
]. The proteomic profile in response to the treatment with CIGB-300, analyzed by 2DE and 2D-LC-MS/MS, revealed, in particular, the instability of nucleolar phosphoprotein B23 (B23/NPM). The effect of this drug was, in fact, to induce the partial degradation of B23 and to impair the ribosomal biogenesis and translation process. These data confirmed that this cyclic peptide may lead to apoptosis by blocking the CK2-mediated phosphorylation on B23/NPM with subsequent interference with the nucleolar assembly. CIGB-300 was also able to modulate other proteins related to metastasis, including thymidine phosphorylase (TP).
As shown in a couple of recent articles [57
], also NSCLC subtypes have been used for the discovery of putative lung cancer biomarkers.
Planque et al.
] have worked on NSCLC cell lines of differing origin, i.e
., (i) adenocarcinoma, H23; (ii) squamous cell carcinoma, H520; (iii) large cell carcinoma, H460; and (iv) small cells, H1688. To avoid complexity and to have more chances for lung cancer biomarker discovery, cells were grown in serum-free medium. From among the proteins identified, tumor necrosis factor-α-converting enzyme (ADAM-17), soluble tumor necrosis factor-receptor type I (sTNF RI), pentraxin 3; osteoprotegerin and follistatin, whose level was higher in the NSCLC of patients compared to controls, were considered putative new lung cancer biomarkers, also on the basis of the fact that each of them had been previously associated with other types of human cancers.
By analyzing the proteome of the conditioned media of QU-BD and Mehr-80, two subtypes of NSCLC that had never been investigated before, Yousefi et al.
] found that stathmin, vimentin, epidermal fatty acid-binding protein, IL-25, transgelin-2, chloride intracellular channel 4 (CLIC4) and stress-induced phosphoprotein I were overexpressed in lung cancer. These proteins should be further investigated as possible biomarkers of large cell lung carcinoma (LCC).
The aim of the following research was to clarify the paracrine mechanisms involved in the crosstalk between human adipose tissue-derived mesenchymal stem cells (hASCs) and cancer cells. Shin et al.
] characterized the secretome of lysophosphatidic acid (LPA)-conditioned medium by shotgun proteomics. TGF β-induced protein ig-h3 (β ig-h3) was shown, for the first time, to play a pivotal role as an extracellular adhesion molecule, which stimulates the proliferation of A549 lung adenocarcinoma cells in vitro.
A variety of extracellular proteins, including periostin, IL-8, insulin-like growth factor-binding protein 3/6 and several proteases/protease inhibitors were also identified as LPA-induced secreted proteins.
The fact that almost all body cells communicate (either directly or through tissues/biological fluids) with the plasma and release (upon damage or death) at least a part of their contents into the bloodstream makes human plasma one of the most important sources of information for clinical proteomics. However, given the complexity of this matrix and the wide range of protein concentrations in plasma, only comprehensive systems characterized by a high resolution and wide dynamic range can be applied to achieve a large-scale analysis of plasma proteome. A variety of attractive procedures [60
] that could revolutionize current methods for the discovery of new disease-associated protein markers are presented below.
Fujii et al.
] applied a µLC system coupled to a linear ion-trap mass spectrometer (µLC-2-D ITMS) with a high scan speed to plasma samples (human serum albumin- and Immunoglobulin G-depleted) from healthy subjects and lung adenocarcinoma patients. Low-abundance proteins, of great clinical importance, because they are directly correlated with the progress of various diseases, could be identified.
An original method consisting in intact protein fractionation followed by chromatographic/ electrophoretic analysis of fractions was developed by Faca et al.
] to achieve in-depth analysis of serum and plasma proteomes. In a few words, pooled serum samples (from healthy controls and newly diagnosed subjects with lung cancer) were depleted of abundant proteins and fractionated by a 2-D LC system consisting of anion-exchange and reversed-phase chromatography. Each fraction was divided into two aliquots, one of which was submitted to shotgun LC-MS/MS and another further resolved on SDS-PAGE. In their turn, the gel bands from this latter step were digested with trypsin and analyzed by MS. Based on the total number of proteins identified and on the representation of specific proteins in individual fractions, they could demonstrate that increased sample fractionation resulted in an increased depth of analysis
To study the altered pattern of protein glycosylation in cancer, Zeng et al.
] developed a label-free method for a glycoproteome analysis of a large set of NSCLC pooled sera that included adenocarcinoma, squamous cell carcinoma) and matched control sera. Their method, based on glycoprotein capture and enrichment, allowed for the isolation of N
-linked glycosylated peptides and their identification by LC-MS/MS. This comparative analysis led to a near complete separation of case and control pools. Further application of this method on a similar set of pooled NSCLC sera led the same authors [63
] to estimate, via spectral counting, the relative abundance of proteins across these pools. More inflammatory response-related proteins were found to be differentially abundant in adenocarcinoma case pools compared to squamous cell carcinoma pools. In their turn, these latter showed a greater response in plasma lipid physiology with more differentially abundant proteins involved in molecular transport.
Applying the same label-free nano-LC-MS/MS method indicated above, Ueda et al.
] obtained a comprehensive peptidome profiling of lung adenocarcinoma serum from fractions that had been enriched by one-step size exclusion chromatography (Mr
range: 1,000–5,000). This approach allowed for the identification of 12 serum peptides as reliable candidate biomarkers for both early detection and tumor staging of lung cancer. Eight of them were fragments derived from fibrinopeptide A, the product of fibrinogen α (FIBA) cleavage; two were generated from apolipoprotein APOA4; one was a fragment from limbin (LBN) and another from amiloride-sensitive cation channel 4 (ACCN 4).
The newly-developed method of Torsetnes et al.
] (that combined selective sample preparation by immunoextraction and identification of signature peptides by LC-MS) was applied to serum samples of SCLC patients. Total progastrin-releasing peptide (ProGRP) was identified as a marker in sera at clinically relevant levels. This method also proved ProGRP isoform occurrence in these sera and differentiated, for the first time, between isoform 1 and 3. From the values of these variants, the amount of ProGRP isoform 2 could also be indirectly calculated.
The attractive method proposed by Oh et al.
] consisted in the combination of longitudinal proteomic analysis with a novel graph-based computational method. In brief, the approach utilizes longitudinal changes of candidate proteins that are located in close proximity or share similar signaling pathways with other biomarkers implicated in the disease that may not necessarily be robust enough for clinical practice. By measuring the closeness between candidate proteins and regularization proteins identified from prior knowledge of the disease process, also a search from a limited sample size may produce relevant biomarkers.
This approach was applied to 26 serum samples collected longitudinally before and during the course of fractionated irradiation treatment of matched-control locally advanced NSCLC patients with and without clinically proven radiation pneumonitis (RP), the manifestation of radiation-induced lung injury. On the basis of the proposed methodology, α-2-macroglobulin (α-2M) was unambiguously ranked as the top candidate protein potentially predictive of early RP onset.
The involvement of serum amyloid A (SAA) in cancer pathogenesis was investigated by Sung et al.
] on crude serum and plasma samples of lung adenocarcinoma and lung cancer of other histological types. Samples were fractionated by SDS-PAGE, gel bands analyzed by LC-ESI-MS/MS and both SAA1 and SAA2 isomers identified. The finding of the higher concentrations in the blood of lung cancer patients than in subjects with other respiratory diseases or other cancers suggested the SAA level in some lung cancers to be a useful differential diagnostic marker. Interestingly, co-culturing lung cancer cells with macrophages resulted in increases of IL-1β and IL-6, which, in turn, stimulate lung cancer cells to induce SAA1/2 production. In conclusion, a lung cancer cell itself expresses and secretes SAA1/2 upon interaction with and stimulation by immune cells residing in the tumor microenvironment. These proteomic data were validated by ELISA analysis.
A linear ion trap quadrupole (LTQ)-Orbitrap platform was applied by Qin et al.
] to pre-diagnostic sera as a new approach to determine whether a set of antigens (consisting of annexin I, P-glycoprotein 9.5 and 14-3-3 theta), previously found to be associated with autoantibodies at the time of diagnosis, discriminates between cases and controls before the onset of symptoms. The occurrence in lung cancer sera of the autoantibodies to antigens defined above and the discovery of laminin receptor-like 1 as a novel lung cancer antigen suggested the potential utility of this approach to diagnose lung cancer before the onset of symptoms. The report by Yildiz et al.
] illustrates a matrix-assisted laser desorption ionization (MALDI)-MS procedure to identify a proteomic signature, directly obtained from unfractionated serum, aimed at distinguishing lung cancer cases from matched controls. From a blinded test set of matched samples, a serum protein signature consisting of seven peptides was found to be associated with lung cancer.
In pursuit of reports dealing with the proteomics of tissues for the biomarker discovery of lung cancer, the article by Marko-Varga et al.
] is worth reading, because it illustrates, with great accuracy, the principles and concepts of different approaches for the discovery of biomarker candidates. Both qualitative and quantitative techniques for determining the specific patterns of proteins detected in tissue and/or biofluids of experimental animal models are described. Human respiratory tract following life-long cigarette smoking in the context of the development of COPD and/or cancer is also documented.
The technical feasibility of a shotgun LC/MS-based method for the global proteomic study of formalin-fixed paraffin-embedded (FFPE) materials was demonstrated by Kawamura et al.
]. Archived clinical FFPE specimens of seven patients with stage IA lung adenocarcinoma without lymph node involvement and another six more advanced stage IIIA subjects, with the spread to lymph nodes, were analyzed to characterize protein expression that could reflect the clinical stages of the disorder. The reliability of this technique was confirmed by multiple-reaction monitoring MS on a subset of these proteins [72
iTRAQ labeling, described above, and the HPLC-Orbitrap-MS platform were used by Li et al.
] to study the expression of tumor-associated proteins and their potential roles in lung adenocarcinomas. All tumors and tumor-matched normal lung tissues used for their investigation were fixed in formalin and embedded in paraffin prior to proteomic analysis. MUC5B, a high-molecular weight, heavily-glycosylated protein, showed significant changes in tumor tissues. The validation of its aberrant expression using immunohistochemistry suggested for this protein a role as a potential biomarker in the detection of adenocarcinomas.
Wei et al.
] developed a new platform to define the proteomic profiles of NSCLC xenografts using a set of ten human tumors [five adenocarcinoma (ADC) and five squamous cell carcinoma (SCC)] that were directly introduced into severely immune deficient mice. The identity and quantity of proteins in the resulting xenograft tumors were obtained from proteomic analysis (in addition to standard histology and immune-histochemistry). Although this study was not aimed at identifying differentially expressed proteins, a comprehensive panel of intermediate filament keratin proteins that could represent a distinctive proteomic signature associated with the NSCLC subtypes was observed. These results confirmed the potential of the method to discover and link tumor molecular markers with the cancer phenotype.
With the aim of searching biomarkers for the early detection of lung squamous cell carcinoma (LSCC), Zeng et al.
] applied iTRAQ in combination with 2D-LC-MS/MS to identify differential proteins among different types of tissues (normal bronchial epithelium; squamous metaplasia; atypical hyperplasia; carcinoma in situ
and invasive LSCC). The combination of glutathione S-transferase P1, heath shock protein beta-1 and creatine kinase brain-type was found to perfectly discriminate normal bronchial epithelium from neoplastic/preneoplastic lesions. For the first time, these three proteins were shown to be novel potential biomarkers for the early detection of LSCC.
Kikuchi et al.
] were the first to use a standardized label-free shotgun proteomic analysis for in-depth tissue protein profiling of the two major subtypes of NSCLC and normal lung tissues. From the analysis of pooled human samples of SCC, adenocarcinoma and control specimens, they identified new pathways and new differentially expressed proteins (i.e
., the p21 activated kinases) of potential interest as diagnostic biomarkers. Multiple reaction monitoring (MRM)-MS confirmed the upregulation of these proteins.
An interesting method (consisting in the combination of 2DE and highly sensitive differential gel electrophoresis (DIGE) saturation labeling for the analysis of limited amounts of microdissected material), established by Poschmann et al.
] to monitor protein changes in SCC tumor progression, revealed that heat-shock protein 47 (HSP47) and a group of cytokeratins were significantly co-regulated in SCC.